This document lays out results for estimating statistical power and minimum detectable effect size (MDES) for: Blocked RCT, with 3 levels, and randomization done at level 2 (school level). This document contains models assuming fixed treatment effects and random treatment effects.

Power Validation

Default parameters:

  • M = 3
  • J = 20
  • K = 10
  • rho: \(\rho\) = 0.5
  • ATE ES = 0.125, 0.125, 0.125
  • R2: \(R^2_1\) = 0, 0, 0, \(R^2_2\) = 0, 0, 0, \(R^2_3\) = 0, 0, 0
  • ICC: \(\text{ICC}_2\) = 0.2, 0.2, 0.2, \(\text{ICC}_3\) = 0.2, 0.2, 0.2
  • Omega: \(\omega_2\) = 0, \(\omega_3\) = 0.1

Varying school size

\(\bar{n}\) = 50

\(\bar{n}\) = 50

\(\bar{n}\) = 50

Varying R2

\(R^2_1 =\) 0.6, 0.6, 0.6

\(R^2_2 =\) 0.6, 0.6, 0.6

\(R^2_3 =\) 0.6, 0.6, 0.6

Varying rho

\(\rho =\) 0.2

\(\rho =\) 0.8

Varying true positives

ATE (ES) = 0.125, 0, 0

Varying ICC

\(\text{ICC}_2\) = 0.7, 0.7, 0.7

\(\text{ICC}_3\) = 0.7, 0.7, 0.7

Varying Omega

\(\omega_3\) = 0.8